hf-tiny-v2/tiny-random-GPTBigCodeForTokenClassification
06
1{2 "activation_function": "relu",3 "add_cross_attention": false,4 "architectures": [5 "GPTBigCodeForTokenClassification"6 ],7 "attention_softmax_in_fp32": false,8 "attn_pdrop": 0.1,9 "bos_token_id": 49152,10 "dtype": "float32",11 "embd_pdrop": 0.1,12 "eos_token_id": 49152,13 "gradient_checkpointing": false,14 "initializer_range": 0.02,15 "layer_norm_epsilon": 1e-05,16 "model_type": "gpt_bigcode",17 "multi_query": true,18 "n_embd": 32,19 "n_head": 4,20 "n_inner": 37,21 "n_layer": 2,22 "n_positions": 512,23 "num_key_value_heads": 1,24 "pad_token_id": 49154,25 "reorder_and_upcast_attn": false,26 "resid_pdrop": 0.1,27 "scale_attention_softmax_in_fp32": false,28 "scale_attn_by_inverse_layer_idx": false,29 "scale_attn_weights": true,30 "tie_word_embeddings": true,31 "transformers_version": "5.16.0.dev0",32 "type_vocab_size": 16,33 "use_cache": true,34 "vocab_size": 4915735}36 